RE: Loop Detected

	From: Tim Ellison/OTT/OTI [mailto:Tim_Ellison@oti.com]

	I think the simplest solution is to say that the server answers the
	property values in a Loop Detected propstat.  This is probably
simpler for
	the server anyway (not having to strip out values because it is a
loop
	detected status).

If the server is going to generate a "Duplicate Detected" status, I don't
think
it would be any problem for it to stuff a DAV:urn in there instead of the
properties list.  If the client cares what those properties are, it can use
the DAV:urn value to find them in the PROPFIND result body.  I think that it
is important that binding not introduce unnecessary duplication of property
values
in PROPFIND responses.

	A note in the spec can point out that if clients want to reconstruct
the
	graph they can deep propfind the dav:urn.

Yes, but that doesn't help them avoid having to parse all those redundant
property
entries in the PROPFIND response.

	I don't think it is a good idea to return a property the client
didn't ask
	for.

I agree that we shouldn't force a server to return DAV:urn for normal status
responses, but I think it reasonable to require the DAV:urn appear for the
"Duplicate Detected" status responses.

Cheers,
Geoff 


	

	                    "Slein, Judith A"

	                    <JSlein@crt.xerox.        To:     "'Clemm,
Geoff'" <gclemm@rational.com>,           
	                    com>                      w3c-dist-auth@w3.org

	                    Sent by:                  cc:

	                    w3c-dist-auth-requ        Subject:     RE: Loop
Detected                            
	                    est@w3.org

	

	

	                    16-03-00 09:35 AM

	

	




	We can make it that the property values get returned for the 506
resource,
	but as you say that will not in general be helpful to the client in
	reconstructing the graph.  It would just be a matter of luck if the
	properties requested allowed you to identify the resource bound to
the
	href.
	As you say, the only property that allows this is DAV:urn.

	So it's not just a matter of changing the example to one where
DAV:urn was
	requested.

	I know you will hate this suggestion, but we could have servers
always
	return DAV:urn, whether it was requested or not, for all the
resources if
	there is a 506 anywhere in the response.

	Or we could add a note suggesting to clients that if they want to
	reconstruct the graph, they should submit another PROPFIND Depth:
infinity
	to the same resource, requesting DAV:urn.

	-----Original Message-----
	From: Clemm, Geoff [mailto:gclemm@Rational.Com]
	Sent: Wednesday, March 15, 2000 4:29 PM
	To: w3c-dist-auth@w3.org
	Subject: RE: Loop Detected


	I agree with Tim.  Furthermore, I would suggest that the example use
	the DAV:urn property (aka DAV:resource-id), since that illustrates
the
	value of the DAV:urn property (the DAV:display-name is not a
reliable way
	of identifying a resource).

	Cheers,
	Geoff

	-----Original Message-----
	From: Tim Ellison/OTT/OTI [mailto:Tim_Ellison@oti.com]
	Sent: Wednesday, March 15, 2000 3:29 PM
	To: w3c-dist-auth@w3.org
	Subject: Loop Detected


	The example given in section 12.1 of
draft-ietf-webdav-binding-protocol-02
	of a PROPFIND depth infinity in the presence of a loop implies that
the
	propstat for the loop detected resource only contains the property
names
	and not their values (again).
	Although I can understand the point of this in the majority of
cases, it
	does prevent the client from reconstructing the graph of resources
since
	they cannot determine the destination of the binding.  If the
properties
	were returned the client could PROPFIND depth infinity on the
resource
	identifier and reconstruct the graph.

	Tim

Received on Thursday, 16 March 2000 11:30:49 UTC